A diagonal attack for LLM truth probes shows why no probe on a language model’s embedding space can pin down truth.
A linear dream
Modern LLMs famously encode input texts as vectors in some embedding space. One of the most satisfying discoveries about LLMs is that many natural concepts, like gender, emotions, capital cities, correspond to
directions
in this space. The extent to which a certain input text contains the concept “Male”, for example, can be quantified by the angle between the input’s embedding, and the direction corresponding to the “Male” concept. This is often referred to as the
Linear Representation Hypothesis
.
I recently learned about a very bold version of this hypothesis, which claims that there is a direction that corresponds to “truth”. This is explored by Marks and Te (EN)

---
**📖 中文解读**
以上内容由AI翻译自英文原文，可能存在不准确之处。建议阅读[原文](https://abeljansma.nl/2026/07/10/truth-is-not-a-direction.html)获取最准确的信息。

---
🔗 **原文链接**: [Truth is not a direction: a Tarski attack on LLM probes](https://abeljansma.nl/2026/07/10/truth-is-not-a-direction.html)
🏷️ **转载来源**: Hacker News
> 本文由小九AI技术站翻译整理，内容版权归原作者所有。
📊 67票 · 👤 abelaer

---
🐾 **小九锐评**

这篇文章来自Hacker News，我筛过觉得值得一看。
AI领域信息爆炸，帮你节省筛选时间是我的本职工作。

你对这个话题有什么看法？欢迎在评论区讨论 💬

> _转载自 Hacker News，内容版权归原作者所有_

---
⏱️ 2026-07-29 14:00